G7: Three AI Lab Chiefs Appear Together
Altman, Amodei and Hassabis face world leaders in Evian-les-Bains. A package of voluntary commitments is on the table, with child safety at the top of the agenda.
- The G7 summit in Evian-les-Bains runs from June 15 to June 17, and for the first time the heads of three leading AI labs are attending.
- Sam Altman, Dario Amodei and Demis Hassabis have confirmed their attendance, and President Macron personally invited Altman.
- A package of voluntary commitments is expected, with child safety at the top of the agenda.
- Risks from advanced AI in the cyber and biological domains are also on the table.
Can the three people who, in practice, decide where artificial intelligence is heading sit down at the same table as world leaders? As of today, we know they can. The G7 summit in Evian-les-Bains, France, opens on June 15 and runs through June 17, and this time the headline news is not economic but technological.
Sam Altman of OpenAI, Dario Amodei of Anthropic and Demis Hassabis of Google DeepMind have all confirmed their attendance. It is the first time the three of them appear together before the leaders of the most developed Western economies. Altman, according to reports, was personally invited by French President Emmanuel Macron, and it is his first appearance at a summit of this kind.
What exactly is being asked of them?
The point is not a front-page photo but a concrete deal. On the table is a package of voluntary commitments, meaning rules the companies accept themselves rather than waiting for every country to pass its own law. That is faster than traditional regulation, but it has an obvious weakness: voluntary means someone can also walk away.
At the top of the agenda is child safety. That covers protecting minors from inappropriate content generated by these models, clearer limits on what systems may do once they detect they are talking to a child, and age checks that actually work. A parent who lets a child use a popular AI tool today usually has no real control over what comes back. That is exactly what G7 leaders want to change.
The second major block of topics covers risks from advanced AI in the cyber and biological domains. In plain terms: how realistic it is for someone to misuse a powerful model to break into systems or to get help producing dangerous materials. The companies already run internal checks for this, but leaders want a shared standard rather than a situation where each lab writes its own exam.
A working lunch is also planned, bringing political leaders and tech executives to the same table. The format looks informal, but the point is serious: for the first time, the people who build the models and the people who write the laws are talking directly, without intermediaries and without scripted statements.
Why does this matter to us too?
It is easy to dismiss this as a topic for Brussels and Washington. But rules agreed at a summit like this tend to become the standard later adopted by the platforms everyone uses. If large models get stricter age verification or clearer limits for minors tomorrow, that will be felt on a teenager’s phone in Belgrade too, not only in Paris.
It is also worth watching who is not at the table. The G7 brings together Western economies, while a large share of AI development happens outside that circle. Voluntary commitments from three companies sound good, but the question is how much they are worth if the rest of the world does not follow the same rules.
So the realistic position sits somewhere between the two extremes. This is neither an empty ceremony nor a moment where the problem is solved. It is the first time the people deciding on artificial intelligence and the people building it are in the same room with a concrete task. What actually survives from these talks will show in whether the voluntary commitments hold the first time they get in someone’s way.
And what worries you more about these everyday tools, protecting children or who actually writes the rules?
